The Pyruvate-Ferredoxin Oxidoreductase−Ferredoxin System

Risques et Nuisances Technologiques-EA 1784, 27 bd. Jean Moulin, 13385 Marseille Cedex 05, France. ReceiVed July 5, 2004; ReVised Manuscript ReceiVed ...
1 downloads 0 Views 940KB Size
Download PDF
15480

Biochemistry 2004, 43, 15480-15493

Multiple Orientations in a Physiological Complex: The Pyruvate-Ferredoxin Oxidoreductase-Ferredoxin System‡ Laetitia Pieulle,§ Matthieu Nouailler,§ Xavier Morelli,§ Christine Cavazza,§,| Philippe Gallice,⊥ Ste´phane Blanchet,§ Pierre Bianco,§ Franc¸ oise Guerlesquin,§ and E. Claude Hatchikian*,§ Unite´ de Bioe´ nerge´ tique et Inge´ nie´ rie des Prote´ ines, Institut de Biologie Structurale et Microbiologie, CNRS, 31 Chemin Joseph-Aiguier, 13402 Marseille Cedex 20, France, and Faculte´ de Pharmacie, Laboratoire de Pre´ Vention des Risques et Nuisances Technologiques-EA 1784, 27 bd. Jean Moulin, 13385 Marseille Cedex 05, France ReceiVed July 5, 2004; ReVised Manuscript ReceiVed September 28, 2004

ABSTRACT: Ferredoxin I from DesulfoVibrio africanus (Da FdI) is a small acidic [4Fe-4S] cluster protein that exchanges electrons with pyruvate-ferredoxin oxidoreductase (PFOR), a key enzyme in the energy metabolism of anaerobes. The thermodynamic properties and the electron transfer between PFOR and either native or mutated FdI have been investigated by microcalorimetry and steady-state kinetics, respectively. The association constant of the PFOR-FdI complex is 3.85 × 105 M-1, and the binding affinity has been found to be highly sensitive to ionic strength, suggesting the involvement of electrostatic forces in formation of the complex. Surprisingly, the punctual or combined neutralizations of carboxylate residues surrounding the [4Fe-4S] cluster slightly affect the PFOR-FdI interaction. Furthermore, hydrophobic residues around the cluster do not seem to be crucial for the PFOR-FdI system activity; however, some of them play an important role in the stability of the FeS cluster. NMR restrained docking associated with site-directed mutagenesis studies suggested the presence of various interacting sites on Da FdI. The modification of additional acidic residues at the interacting interface, generating a FdI pentamutant, evidenced at least two distinct FdI binding sites facing the distal [4Fe-4S] cluster of the PFOR. We also used a set of various small acidic partners to investigate the specificity of PFOR toward redox partners. The remarkable flexibility of the PFOR-FdI system supports the idea that the specificity of the physiological complex has probably been “sacrificed” to improve the turnover rate and thus the efficiency of bacterial electron transfer.

Oxidative decarboxylation of pyruvate to form acetylcoenzyme A is a crucial step in cell metabolism. In aerobic organisms, this reaction is catalyzed by the pyruvate dehydrogenase multienzyme complex (PDH).1 In most anaerobic bacteria, the reaction is carried out by the pyruvate-ferredoxin oxidoreductase (PFOR). The electron acceptor of this single enzyme is either a [4Fe-4S] ferredoxin or a flavodoxin. Although both PFOR and PDH are thiamin pyrophosphate ‡ Nucleotide sequence data are available in the DDBJ, EMBL, and GenBank databases under accession number AJ489482. * To whom correspondence should be addressed: Unite´ de Bioe´nerge´tique et Inge´nie´rie des Prote´ines, IBSM, CNRS, 31, Chemin JosephAiguier, 13402 Marseille Cedex 20, France. Telephone: 33 4 91 16 41 45. Fax: 33 4 91 16 45 78. E-mail: [email protected]. § CNRS. | Present address: Laboratoire de Cristallographie et Cristallogene`se des Prote´ines, Institut de Biologie Structurale Jean-Pierre Ebel, Commissariat a` l’Energie Atomique, Universite´ Joseph Fourier, CNRS, 41, Jules Horowitz, 38027 Grenoble Cedex 1, France. ⊥ Laboratoire de Pre´vention des Risques et Nuisances TechnologiquesEA 1784. 1 Abbreviations: PFOR, pyruvate-ferredoxin oxidoreductase; PDH, pyruvate dehydrogenase multienzyme complex; TPP, thiamin pyrophosphate; Da, DesulfoVibrio africanus; Fds, ferredoxins; Dg, DesulfoVibrio gigas; Tc, Thermodesulfobacterium commune; Dmn, Desulfomicrobium norVegicum; Cp, Clostridium pasteurianum; Ssp., Spirulina sp.; FdI, ferredoxin I; FdII, ferredoxin II; htFdI, His-tagged ferredoxin I; IPTG, isopropyl β-D-thiogalactopyranoside; DTE, dithioerythritol; ITC, isothermal titration calorimetry; PCR, polymerase chain reaction; HSQC, heteronuclear single-quantum correlation.

(TPP)-containing enzymes, only the former can carry out the reverse reaction, that is, the reductive carboxylation of acetyl-CoA to yield pyruvate. This reaction constitutes the basis for CO2 fixation in green photosynthetic and acetogenic bacteria (1) and in methanogens (2). PFORs are homodimeric, heterodimeric, or heterotetrameric enzymes that are phylogenetically related. The structure of DesulfoVibrio africanus (Da) PFOR has been determined by X-ray crystallography (3). This homodimeric enzyme contains one TPP molecule per subunit, buried in the protein, and three [4Fe-4S]2+/1+ clusters with midpoint potentials ranging from -540 to -390 mV (4). These redox centers are regularly disposed between the active site and the molecular surface, providing a potential electron transfer pathway (5). In Da, three ferredoxins (Fds) have been isolated and characterized (6); the monocluster ferredoxin I (FdI), the most abundant, is a small protein (7.5 kDa) containing one [4Fe-4S]2+/1+ cluster with a low redox potential (E°′ ) -385 mV), and its structure has been determined by both X-ray crystallography (7) and NMR (8). Kinetic parameters of the electron transfer between PFOR and FdI from Da have been previously determined by electrochemistry (6). The second-order rate constant is 7 × 107 M-1 s-1, and the Km value of the enzyme for FdI is 3 µM. The fast electron transfer between Da PFOR and FdI suggests that a transient complex is formed. Cross-linking experiments between PFOR and FdI have shown that the

10.1021/bi0485878 CCC: $27.50 © 2004 American Chemical Society Published on Web 11/17/2004

Pyruvate-Ferredoxin Oxidoreductase-Ferredoxin I Complex

Biochemistry, Vol. 43, No. 49, 2004 15481

Table 1: Template and Oligonucleotide Primer DNA Sequence mutant

template

oligonucleotidea

E15A D7A/D9A/E10A ∆62-64 D7A/D9A/E10A/ ∆62-64 D7A/D9A/E10A/ D28N/E30Q/E32Q D52R D52A/E48A/E49A

pLF1 pLF1 pLF1 pLF1var.D7A/D9A/E10A

(+200stop+cattgcctgcgCGtcctgcgtggagatcgcacc)+(-200pTac+cacgcaggaCGcgcaggcaatgcattcg) taggtacccaaggaggcatcaatg(cac)6gctcgcaagttctatgtaGCGcagGCGGCGtgcattgcc+FdIBamHIb ttggatccgcctaccaatggatgctctgcacagg+FdIKpnIb ttggatccgcctaccaatggatgctctgcacagg+FdIKpnIb

pLF1var.D7A/D9A/E10A

D52A/ E45A/E46A/ E48A/E49A I12T I12E I20E Y35E M27E

pLP1

(+200stop+tgccatgAacccgCagattCagaaggcctatgtc)+(-200pTac+ggccttctGaatctGcgggtTcatggcaaaggcg) (+200stop+gtcgaggaggccatgCGtacctgccctgtg)+(-200pTac+cacagggcaggtaCGcatggcctcctcgac) (+200stop+ccaggaagaagtcgCTgCggccatggCAacctgccctgtg)+(-200pTac+cacagggcaggtTGccatggccGcAGcgacttcttcctgg) (+200stop+gcgccagccaggCTgCagtcgCTgCggccatggCAacctgccctgtg)+(-200pTac+cacagggcaggtTGccatggccGcAGcgactGcAGcctggctggcgc) taggtacccaaggaggcatcaatg(cac)6gctcgcaagttctatgtagaccaggacgaatgcACCgcctgcg+FdIBamHIb taggtacccaaggaggcatcaatg(cac)6gctcgcaagttctatgtagaccaggacgaatgcGAAgcctgcg+FdIBamHIb (+200stop+tcctgcgtggagGAAgcaccgggcgcc)+(-200pTac+ccggtgcTTCctccacgcaggattcgc) (+200stop+gattgagaaggccGAAgtcaaggacgtg)+(-200pTac+cgtccttgacTTCggccttctcaatctcc) (+200stop+gcctttgccGAAgacccggagattgagaagg)+(-200pTac+ctccgggtcTTCggcaaaggcgcc)

pLF1 pLF1

pLF1 pLF1 pLF1 pLF1 pLF1

a Change made to the native sequence at the position indicated in bold capital letters. b Described in Experimental Procedures: +200 stop primer, 5′-gcgtttcacttctgagttcgg-3′; -200 pTac primer, 5′-gcagaaacgtggctggcctgg-3′. The mutagenic oligonucleotides are written in the usual 5′-3′ direction.

covalent complex exhibits a stoichiometry of one FdI for one PFOR monomer (6). The formation of the cross-linked complex and the electron transfer between the two redox partners are sensitive to the ionic strength, suggesting that electrostatic interactions are the driving force in formation of the complex (6). The molecular surface in the vicinity of the distal [4Fe-4S]2+/1+ cluster of PFOR displays several positively charged conserved residues that constitute a potential recognition site for the negatively charged surface of FdI (3). Electron transfer between proteins generally requires formation of a transient complex that brings together the two redox centers exchanging electrons to achieve a high turnover rate. It is generally assumed that the formation of the complex takes place in two stages, including (i) the initial encounter dominated by relatively long-range electrostatic forces and (ii) the second state involving additional short-range forces that include hydrogen bonds, van der Waals forces, and hydrophobic interactions; these make important contributions both to the overall binding energy and to the specificity of the interaction (9). This study was designed to extend our knowledge of the structural features that are involved in the binding and subsequent electron transfer between PFOR and FdI. The cocrystallization of the PFOR-FdI complex is a real challenge; an alternative to obtaining a structural model of the complex is NMR-restrained docking (10) associated with site-directed mutagenesis data. With this aim in mind, the cDNA encoding Da FdI was cloned and sequenced and the recombinant FdI was expressed in Escherichia coli. In an effort to evaluate the role of electrostatic forces in formation of the complex, most of the FdI negatively charged surface residues (15 of 18) were then substituted with either a neutral or a charge-reverse residue or deleted. Therefore, a panel of FdI molecular variants, including one, three, five, or six carboxylate residues replaced simultaneously, was generated. Thermodynamic parameters of formation of the complex were analyzed through microcalorimetry experiments, and the electron transfer efficiency was characterized by steady-state kinetics. NMR-restrained docking was used to clarify the effects induced by carboxy-

late residue substitutions and suggested a new set of mutations. EXPERIMENTAL PROCEDURES Bacterial Strain E. coli strain TG1 [supE hsd∆5 thi∆(lac-proAB) F′[traD36 proAB+ lacIq lacZ] ∆M15] was used for Da FdI expression. Cloning and Sequencing of the fd1 Gene Cloning of the fd1 gene was carried out as previously described (11). The degenerate oligonucleotides [5′-TAY GTN GAY CAR GAY GAR TG-3′] and [5′-TCR TCY TCC CAR TGD ATR CA-3′] were derived from the amino acid sequence of Da FdI (7). The PCR product was cloned directly into the pCR2.1 vector using the TA cloning kit (Invitrogen) and sequenced. Two outward-facing oligonucleotides, [5′CTT CTC AAT CTC CGG GTC CAT GGC-3′] for the 5′extremity of the fd1 gene and [5′-AGC CAG GAA GAA GTC GAG GAG GCC-3′] for the 3′-extremity of the fd1 gene, were used for inverse PCR. Site-Directed Mutagenesis of Da FdI The fd1 gene was amplified by PCR using oligonucleotides FdIKpnI and FdIBamHI as primers. The sequence of FdIKpnI [5′-TAG GTA CCC AAG GAG GCA TCA ATG (CAC)6 GCT CGC-3′] is composed of a KpnI recognition site (underlined in the sequence) followed by the putative ribosome binding site of the fd1 gene (boldface in the sequence), and then 20 nucleotides corresponding to the beginning of the fd1 coding sequence, including the ATG and six triplets (boldface) encoding the His-tagged domain for convenient purification. The sequence of FdIBamHI [5′T TGG ATC CGC CTA CTC GTC CTC CCA ATG GAT GC-3′] contains a BamHI cleavage site (underlined) followed by two nucleotides, a triplet complementary to the stop codon (boldface), and nucleotides complementary to the 3′-end of the fd1 coding sequence. The KpnI-BamHI fragment was introduced into the polylinker cloning site of plasmid pJF119EH (11) to give plasmid pLF1. FdI mutants were

15482 Biochemistry, Vol. 43, No. 49, 2004 prepared from plasmid pLF1 using a PCR-based procedure. Primers for mutagenesis are listed in Table 1. The fd1 gene was sequenced to verify the mutations and to check for any mistakes that might have occurred elsewhere. Expression and Purification of Da Recombinant FdI and Molecular Variants Recombinant FdI and molecular variants were expressed in E. coli. Cells were grown aerobically at 37 °C in a modified Terrific Broth medium containing 12 g of yeast extract (Difco), 10 g of bacto-tryptone (Difco), 12.5 g of K2HPO4, 2.3 g of KH2PO4, 8 mL of glycerol, and 1000 mL of distilled water. Before inoculation, the medium was supplemented with 100 mg/L ferric ammonium citrate according to the method of ref 12 and 100 mg/L ampicillin. Large-scale cultures were grown in a 10 L fermentor (Chemap) while being stirred at 400 rpm with an aeration of 10 L/min. When the culture reached 1.3 A600 units, production of FdI was induced for 3 h with 0.5 mM IPTG (isopropyl β-D-thiogalactopyranoside), and the growth medium was supplemented once more with ferric ammonium citrate. The cells (180 g) were harvested by centrifugation at 50000g in a continuous-flow centrifuge (Sharples) when the growth reached 12 A600 units and stored at -80 °C. The harvested cells were resuspended with 20 mM potassium phosphate (pH 8.0) and 500 mM NaCl (buffer A) containing 1 µM deoxyribonuclease I (Sigma). The cells were passed once through a French pressure cell at 100 MPa, and cell debris was removed by centrifugation at 35000g for 20 min. All further purification steps were carried out under a nitrogen atmosphere. The supernatant was treated with deoxyribonuclease I and ribonuclease A (1 µM each, Sigma) at 30 °C for 30 min and subsequently centrifuged at 120000g for 90 min. The soluble protein fraction which contains Da His-tagged FdI (htFdI) was further purified by affinity chromatography on a Ni-NTA agarose (Qiagen) column. The soluble extract was loaded onto the column (2.6 cm × 1.5 cm) equilibrated with buffer A containing 5 mM imidazole. The column was washed first with 150 mL of the same buffer followed by 40 mL of buffer A containing 15 mM imidazole. Finally, the brown protein band bound at the top of the column was eluted with buffer A containing 80 mM imidazole. The htFdI-containing fraction was subsequently concentrated on a YM10 membrane (Centriplus, Amicon) and purified by gel filtration on a Sephacryl S-100 HR (Pharmacia) column (85 cm × 2.6 cm) equilibrated with 50 mM Tris-HCl buffer (pH 8.5). The Fd fraction was then concentrated on a YM10 membrane and stored in liquid nitrogen. At this stage, Da htFdI was considered to be pure on the basis of both polyacrylamide gel electrophoresis and absorption ratio (A385/A283 ) 0.58) (13). 15 N labeling of htFdI and the FdI pentamutant was carried out by growing E. coli cells in 4 L of minimal medium (M9) supplemented with [15N]ammonium sulfate (1 g/L) and glucose as a carbon source (8 g/L). Cell growth conditions and protein expression were the same as described before, and ferric ammonium citrate was omitted. Proteins and Enzyme Purification PFOR, FdI, and ferredoxin II (FdII) were purified from Da (NCIMB 8401) as previously described (4, 13). The

Pieulle et al. purification procedures for cytochrome c3, FdI and FdII, flavodoxin, and rubredoxin from DesulfoVibrio gigas (Dg) (14) as well as Thermodesulfobacterium commune (Tc) Fd (15) were previously reported. Desulfomicrobium norVegicum (Dmn) FdI and FdII were purified as described previously (16). Clostridium pasteurianum (Cp) Fd and Aquifex aeolicus Fd 1 were gifts from J.-M. Moulis and J. Meyer, respectively (DBMS, CEA, Grenoble, France). Spirulina sp. (Ssp.) Fd was purchased from Sigma. Analytical Methods Gel Electrophoresis. Slab gel electrophoresis in the presence of SDS was performed according to the method of Laemmli (17), with a 15% polyacrylamide running gel. Protein samples were previously heated at 100 °C for 5 min in the presence of 2% SDS and 1 mM mercaptoethanol. Protein bands were visualized by staining with Coomassie blue R250. S-Carboxymethylation and Sequence Determination. htFdI was reduced and S-carboxymethylated as previously reported (18), and N-terminal amino acid sequence determination was performed by stepwise Edman degradation using a model 473A automatic sequencer (Applied Biosystems). UV-Visible Spectroscopy. Absorption spectra were carried out on a model UV 1601 spectrophotometer (Shimadzu). The concentrations of Da PFOR and FdI as well as its molecular variants were determined using molecular extinction coefficients of 105 mM-1 cm-1 at 400 nm and 15 mM-1 cm-1 at 385 nm, respectively. ActiVity Assay The biological activity of htFdI and molecular variants was assessed in a coupled assay system in which FdI reduced by PFOR was subsequently reoxidized by an excess of cytochrome c3. This assay allows us to maintain a constant amount of oxidized FdI in the reaction mixture. Previous data indicate that electron transfer from Fd to cytochrome c3 (19) is faster than that from PFOR to Fd (6). Dg cytochrome c3 was used in the coupled assay because direct reduction of this cytochrome by PFOR is negligible. Assays were performed under anaerobic conditions at 30 °C in rubber-stoppered cuvettes (total volume of 0.5 mL) containing 50 mM Tris-HCl buffer (pH 8.5), 10 mM pyruvate, 0.2 mM CoASH, 0.2-15 µM Da FdI, and 40 µM Dg cytochrome c3. After the cuvette had been bubbled for 20 min with argon, PFOR (2-3 pmol) was injected and the reduction of cytochrome c3 was monitored at 552 nm. Initial rates of cytochrome c3 reduction were calculated using a difference absorption coefficient of 88 mM-1 cm-1 at 552 nm. Under our assay conditions, the oxidation of one molecule of pyruvate catalyzed by PFOR is coupled to the reduction of two molecules of Da FdI, whereas the reduction of one molecule of tetraheme cytochrome c3 corresponds to the reoxidation of four molecules of FdI reduced by PFOR. The kinetic parameters have been adjusted to the hyperbola equation according to the Cleland method (20). For evaluating the reactivity of Da PFOR toward various natural electron carriers, the activity of the enzyme has been determined by measuring the level of reduction of the electron acceptor at 420 nm for ferredoxins, 456 nm for flavodoxin, and 490 nm for rubredoxin. Assays were carried

Pyruvate-Ferredoxin Oxidoreductase-Ferredoxin I Complex out under anaerobic conditions in a serum-stoppered cuvette containing 50 mM Tris-HCl buffer (pH 8.5), 10 mM pyruvate, 0.2 mM CoASH, 20 µM electron carrier, and 16 mM DTE, in a final volume of 0.5 mL. The reaction was started by injection of PFOR (0.4 pmol). Electrochemical Measurements Cyclic (CV) and square-wave (SWV) voltammetry were carried out using an EG&G 273A potentiostat modulated by EG&G PARC M270/250 software. A three-electrode cell consisting of a Metrohm Ag |AgCl| saturated NaCl reference electrode (having a potential of 210 mV vs the normal hydrogen electrode), a gold wire auxiliary electrode, and a working pyrolytic graphite electrode was used throughout. Before each experiment, the working electrode was polished with 0.05 µM alumina slurry. Because of the excess of negative charges of acidic proteins such as htFdI or molecular variants, the rate of electron transfer to or from the working electrode required the concentration of the polycations added to the solution, here 10 µM poly(L-lysine), to be increased (21), and thus the electrode responses to be detected. The reversibility of electron exchanges was established from CV curves, and the redox potential values were calculated from SWV peak potentials. All experiments were performed at room temperature, in 0.01 M Tris-HCl (pH 7.6) under an argon atmosphere. Isothermal Titration Calorimetry Analysis Reagents. Da PFOR and FdI solutions were equilibrated in 10 mM Tris-HCl (pH 8.5) and 8.55 mM NaCl to obtain concentration values close to 10 µM and 3.7 mM, respectively. To obtain PFOR and FdI under identical buffering conditions, the protein solutions were extensively dialyzed and concentrated by ultrafiltration using Centricon-10 concentrators (10 000 MW cutoff, Amicon) for FdI and Centricon-50 concentrators (50 000 MW cutoff, Amicon) for PFOR. Isothermal Titration Calorimetry (ITC). Equilibrium binding experiments designed to study the interaction between PFOR and FdI were performed at 298 K using a 2277 Thermal Activity Monitor calorimeter (Thermometric) equipped with a titration unit. Data acquisition and analyses were carried out using DIGITAM 4.1 (Thermometric). In a typical experiment, titration was performed as follows. Ten aliquots (5 µL each) of the FdI solution were injected from a 100 µL syringe into the calorimeter cell containing the PFOR solution (0.9 mL) to achieve a complete binding isotherm. The effective heat of binding was obtained by subtracting the heat of dilution (measured by additional injections of the FdI solution after saturation) from the heat of reaction. The binding enthalpy (∆H°) and affinity constant (Ka) were then obtained by fitting the experimentally corrected binding isotherm to a model incorporated into DIGITAM, assuming one set of binding sites. Changes in free energy (∆G°) and entropy (∆S°) are obtained by the calculation ∆G° ) -RT ln Ka ) ∆H° - T∆S°. According to the methodology described, ITC was used in a set of experiments to investigate the effect of ionic strength on the interaction. Titrations were thus carried out in 10 mM Tris-HCl (pH 8.5) with 0.015, 0.03, and 0.1 M NaCl.

Biochemistry, Vol. 43, No. 49, 2004 15483 NMR-Restrained Docking NMR. NMR experiments were carried out on a Bruker DRX 500 spectrometer at 298 K. The assignments of 1H and 15N chemical shifts of htFdI and the pentamutant were obtained from three-dimensional NOESY-HSQC and HNHA experiments performed on a 15N-labeled FdI sample at a concentration of 0.7 mM in 20 mM phosphate buffer (pH 6.1) and 10% D2O. Proton chemical shifts were found to be in agreement with previous FdI homonuclear assignments (8). NMR titrations were performed using two-dimensional HSQC experiments, recorded using a Watergate pulse sequence in the TPPI mode, on a 0.06 mM htFdI or pentamutant sample in the absence or presence of various molar equivalents of PFOR (0.1, 0.2, 0.5, and 1). In HSQC experiments, the chemical shift variations are observed on exposed NH groups located at the interacting site, the side chains of the corresponding residues being buried in the molecule (this is the case for hydrophobic residues such as Val or Ile). Residues involved in protein-protein interactions (like acidic residues) have exposed side chains and buried NH groups. The NH resonances of these residues may be not affected by formation of the complex. Thus, chemical shift variations give the mapping of an interacting site. The spectral widths are 2000 Hz for 1H and 2027 Hz for 15N; 1024 data points in t2 and 64 transients for each of 128 t1 points were used. Docking. Ab initio docking calculations were performed using BiGGER version 1.0 (22). The algorithm starts with the atomic coordinates of the two proteins (PDB entry 1FXR for FdI and PDB entry 1B0P for PFOR). For the pentamutant, a model has been generated by residue substitutions in the FdI structure. Dockings were obtained after a systematic search of the binding space of both molecules (10° rotation, 1 Å translation), and the algorithm generates a population of 5000 hypothetical complexes. These solutions are ranked according to a scoring function which combines several interaction terms that are thought to be relevant for the stabilization of protein complexes: geometric packing of surfaces, electrostatic interactions, desolvation energy, and pairwise propensities of the amino acid side chains to make contact across the molecular interface. NMR Filtering. Chemical shift variations induced by complex formation were used to select the best model obtained from ab initio calculation, according to the restrained docking approach (23). A combination of 1H and 15N chemical shift variations was used for NMR filtering. Chemical shift variations of >0.012 ppm are considered to be significant. The chemical shift variations were converted into distances (a nucleus is affected when it is within 5 Å of any atoms of the partner). The representation of the protein structure was obtained using Pymol 0.95 from W. L. DeLano [PyMOL (2002) DeLano Scientific, San Carlos, CA]. RESULTS Characterization of the PFOR-FdI Complex by ITC The Da PFOR-FdI electron transfer complex has been characterized by electrochemistry (6). In the work presented here, we have established two sets of data, the thermodynamic parameters obtained from ITC and electron transfer

15484 Biochemistry, Vol. 43, No. 49, 2004

Pieulle et al.

FIGURE 1: Heats of binding (A) and binding isotherm (B) for the calorimetric titration of 10.7 µM PFOR with 5 µL injections of 3.87 mM FdI (wild-type) in 10 mM Tris-HCl buffer and 8.55 mM NaCl at pH 8.5 and 298 K.

parameters obtained by steady-state kinetics using spectrophotometry. Formation of the PFOR-FdI complex produces an endothermic heat of binding (Figure 1A). A typical calorimetric titration curve is obtained, indicating a stoichiometry of complex formation of two molecules of FdI per dimer of PFOR (Figure 1B). The association constant (Ka) determined from the plot is 3.85 × 105 M-1. The enthalpy change (∆H°) is 21.3 kJ/mol, and the entropy change (∆S°) is 0.178 kJ mol-1 K-1. These values indicate that hydrophobic interactions contribute to complex formation, but are also consistent with electrostatic interactions associated with the release of solvated water molecules. Consequently, it is difficult to identify the driving force for the interaction from thermodynamic data alone (24). The electrostatic interactions were studied by altering the ionic strength of the buffer with NaCl. At a total ionic strength of 100 mM, the binding was too weak to be assessed. An increase in ionic strength from 10 to 30 mM weakens the binding of FdI by a factor 6.5, indicating that electrostatic interactions are essential in complex formation as already suggested by kinetics measurements (6). The increase in ionic strength induces a decrease in the entropic driving force (from 0.178 to 0.147 kJ mol-1 K-1), whereas the enthalpy change becomes more favorable (from 21.3 to 16.5 kJ/mol).

Effects of FdI Substitutions on the PFOR-FdI Complex Cloning, Expression, and Characterization of Da FdI. The gene encoding Da FdI was cloned and sequenced. The fd1 gene nucleotide sequence was submitted to the EMBL/ GenBank (accession number AJ489482) and found to be in agreement with the protein sequence previously reported (7). Moreover, the absence of an N-terminal signal peptide confirmed the presumed cytoplasmic localization of FdI. The fd1 coding sequence was cloned downstream from the IPTGinducible tac promoter of plasmid pJF119EH, and FdI was isolated using a histidine tag at the N-terminus. The purification procedure of htFdI yielded 1.6 mg of pure protein per liter of culture. Amino acid sequencing of the carboxymethylated htFdI showed the presence of a single polypeptide chain with the N-terminal sequence of Da FdI (ARKFYVDQDEC), preceded as expected by the six histidine residues. The redox potential, UV-visible, and EPR spectroscopic features of the htFdI (not shown) are indistinguishable from those of the wild type (13, 25). Kinetic parameters of the reduction of FdI and htFdI were determined (Km ) 0.8 µM and kcat ) 139 s-1). The Km value for FdI measured in the coupled assay with cytochrome c3 is 3-4-fold lower than the values previously obtained using the coupled assay with metronidazol or the electrochemical method (6).

Pyruvate-Ferredoxin Oxidoreductase-Ferredoxin I Complex

Biochemistry, Vol. 43, No. 49, 2004 15485

FIGURE 2: (A) Sequence alignment of Fds from Da, Dmn, Dg, and Cp. Sequences were aligned using Clustal W. The numbering corresponds to the Da FdI sequence. FeS ligand cysteines are colored red. Acidic residues at the interacting surface of FdI and conserved in other Fds are colored green. Hydrophobic residues at the interacting surface and conserved in other Fds are colored blue. Asterisks denote the mutated residues in Da FdI. (B) Structure of Da FdI (PDB entry 1FXR). On the left are represented the side chains of the first set of mutated residues (top part of Tables 2 and 3). On the right are represented the side chains of the second set of mutated residues suggested to be involved in formation of the complex by NMR experiments (bottom part of Tables 2 and 3).

Substitution of Charged Residues. Da FdI contains a great number of anionic residues (18 aspartic or glutamic acid of 64 residues) spread at the surface of the molecule (Figure 2B), and many of them are highly conserved in the sequences of DesulfoVibrio monocluster ferredoxins (Figure 2A). Therefore, seven carboxylates (D7, D9, D28, E10, E15, E30, and E32) surrounding the [4Fe-4S] cluster were selected as primary targets for site-directed mutagenesis (Figure 2B). Many kinds of substitutions involving one, three, and up to six carboxylates at the same time have been performed (Table 2). Moreover, the C-terminal extremity (E62-D63-E64), which forms a highly negative hook pointing out of the FdI molecule proposed to play a role in the recognition of the redox partner (6), has been deleted (Figure 2B). The expression of all the variants was similar to that of htFdI except for the D7A/D9A/E10A/D28N/E30Q/E32Q and D7A/D9A/E10A/∆62-64 mutants which yielded only 30%

of that of the recombinant protein. The values of the midpoint redox potentials of the variants indicate that neutralization of surface acidic residues by site-directed mutagenesis has no appreciable effect on the cluster redox potential (top part of Table 2). These results agree with those previously obtained with Azotobacter Vinelandii FdI (26). The kinetic parameters of the reduction of Da FdI variants by PFOR were compared to those of FdI (top part of Table 2). None of the variants for which one or several carboxylic groups were substituted displayed a change in Vmax/Km larger than 20% of the ratio exhibited by FdI. Moreover, the highly acidic C-terminal extremity of the FdI molecule seems not to be critical for the interaction with PFOR. Calorimetric titration was performed for five variants (Table 3). The instability of the D7A/D9A/E10A/∆62-64 variant has not allowed a precise determination of its Ka. htFdI and the E15A variant exhibit Ka values similar to that

15486 Biochemistry, Vol. 43, No. 49, 2004

Pieulle et al.

Table 2: Comparison of Redox Potentials ((10 mV) and Kinetic Parameters of Da Recombinant FdI and Molecular Variantsa ferredoxin I wild-type htFdI E15A D7A/D9A/E10A ∆62-64d D7A/D9A/E10A/ ∆62-64 D7A/D9A/E10A/ D28N/E30Q/E32Q I12T I12E I20E Y35E D52R E48A/E49A/D52A E45A/E46A/E48A/ E49A/D52A

Emb (mV)

kcat (s-1)

Km (µM)

kcat/Km (%)c

-380 -370 -385 -385 -375 -370

139.3 ( 7.2 138.8 ( 8.7 123.6 ( 17.7 128.0 ( 9.4 134.8 ( 8.4 125.6 ( 9.0

0.81 ( 0.10 0.78 ( 0.13 0.61 ( 0.10 0.85 ( 0.14 0.87 ( 0.15 0.65 ( 0.12

100 103 118 88 90 112

-400

130.4 ( 17.7

0.71 ( 0.12

107

-360 -380 -400 -400 -400 -400 -410

132.8 ( 8.8 104.4 ( 17.4 130.4 ( 8.7 150.8 ( 8.9 141.7 ( 8.7 126.7 ( 1.7 113.0 ( 3.7

1.23 ( 0.28 1.23 ( 0.02 0.70 ( 0.13 1.00 ( 0.15 1.92 ( 0.28 1.42 ( 0.29 1.57 ( 0.16

63 49 108 88 43 52 42

a

Under the assay conditions, FdI and FdI variants couple Dg cytochrome c3 reduction to PFOR oxidation. The rate of Fd reduction in the coupled assay was determined as described in Experimental Procedures. b Em is the redox potential value vs the normal hydrogen electrode. The concentration of FdI and FdI variants used in the measurement of redox potentials by square-wave voltammetry was in the range of 80-150 µM. c kcat/Km values are given as the percentage of the value for wild-type FdI. d ∆62-64 is the form in which the EDE tripeptide of the C-terminal extremity has been deleted.

of wild-type FdI, whereas for all the other variants, the Ka is slightly diminished. However, some significant variations of binding enthalpy were observed; as an example, the ∆H value of the D7A/D9A/E10A/D28N/E30Q/E32Q variant (∆H° ) 7.3 kJ/mol) is strongly more favorable than that measured with the D7A/D9A/E10A variant (∆H° ) 14.4 kJ/mol), whereas the free energy is the same (∆G° ) -30.2 and -30.0 kJ/mol, respectively). Thus, it appears clearly that the enthalpy change is counteracted by a favorable entropy change. This enthalpy-entropy compensation is observed for the set of variants studied, and the linear relationship obtained between ∆H° and ∆S° provides a slope correlated with the experimental temperature. These results indicate that the variations of enthalpy are probably due to the removal of water molecules from the interacting surface (27). Substitution of Hydrophobic Residues. In addition to the potential role of long-range electrostatic forces in proteinprotein interaction associated with charged residues, shortrange forces, due mainly to surface hydrophobic residues, make important contributions to the overall binding energy and to the specificity of the interaction between redox partners (9, 28, 29). Hydrophobic residues I12, I20, and M27 located at the molecular surface surrounding the [4Fe-4S] cluster of FdI as well as Y35 have been substituted with a negatively charged (E) residue. In addition, I12 has been substituted with a polar (T) residue. Among these molecular FdI variants, M27E, I12E, and I12T were unstable. The UVvisible spectrum of the weakly expressed M27E form obtained after the Ni-NTA column was indicative of oxidative damage of the [4Fe-4S] center. Therefore, this highly unstable variant has not been used in the activity assay. The yield of I12E and I12T mutants was only 30% of the amount of htFdI. The midpoint redox potentials of this set of variants (Table 2) are close to that of htFdI except

for the redox potentials of I20E and Y35E which are slightly negatively shifted. All these singly substituted mutants displayed kinetic properties close to those of FdI except I12E and I12T, which showed 51 and 37% decreases in their catalytic efficiency, respectively (Table 2). However, unlike wild-type FdI, I12E and I12T mutants exhibit a g ) 2.01 [3Fe-4S] cluster EPR signal in the oxidized state, corresponding to approximately 20% of the signal from the reduced protein. The amount of [3Fe-4S] cluster increased over time, which is indicative of the instability of the [4Fe-4S] cluster in these proteins. The instability of I12 variants has not allowed a precise determination of their Ka values. Finally, ITC analysis shows similar effects of I20E and Y35E substitutions as compared to charged residue substitutions (Table 3). Structural Models of the PFOR-FdI Complex through NMR-Restrained Docking Two observations can be made from FdI site-directed mutagenesis experiments. (i) No targeted charged residue has a crucial role in complex formation; this result is in disagreement with the high sensitivity of FdI binding affinity to ionic strength as shown by ITC analysis. (ii) Most hydrophobic residues are essential for the stability of the FdI [4Fe-4S] cluster and thus are not good targets for site-directed mutagenesis analysis. Thus, to understand how PFOR and FdI interact, a structural analysis of the complex was carried out through NMR-restrained docking. NMR Titration of Formation of the PFOR-FdI Complex. The 15N-1H HSQC experiment (Figure 3) is the fingerprint of Da FdI (at 296 K), where each cross-peak corresponds to the correlation between 15N and 1H atoms of the amide groups. The sequential assignment of these peaks was obtained from heteronuclear NMR experiments. Due to the paramagnetic effect of the FeS cluster at 296 K, only 51 well-resolved correlation peaks were identified among the 64 residues of FdI and used for the mapping of the PFOR interacting site. HSQC experiments have been carried out in the presence of various PFOR concentrations (0.1, 0.2, 0.5, and 1 equiv of FdI). The size of the PFOR dimer (270 kDa) induces line broadening of FdI resonances which increases with PFOR concentration, and most of the lines disappeared at 0.5 equiv. The shifts of the resonances indicate that the complex is in fast exchange at the NMR time scale. Among the 51 resonances observed in the FdI HSQC spectrum (Figure 3), eight resonances (V6, C14, E19, I20, I31, G41, S43, and A50) are notably affected by complex formation (Figure 4A). The mapping of the interacting site of PFOR on FdI revealed an extensive surface area affected by complex formation, including acidic residues largely distributed around the protein. Docking of the PFOR-FdI Complex. Ab initio docking of the PFOR-FdI complex was generated by BiGGER using PDB entries 1FXR and 1B0P. Figure 5A presents the top 1000 of the initial 5000 hypothetical complexes, ranked according to BiGGER’s scoring function. The PFOR dimer is represented here, and a relatively good symmetrical distribution of FdI (each ball represents the FdI center of mass) is observed on the two subunits. Most of the good solutions according to the BiGGER scoring function are located near the distal cluster of PFOR at the potential recognition site suggested by the X-ray structure (3).

Pyruvate-Ferredoxin Oxidoreductase-Ferredoxin I Complex

Biochemistry, Vol. 43, No. 49, 2004 15487

Table 3: ITC Analysis of Da FdI and Molecular Variants Binding to PFOR

a

ferredoxin I

Ka (M-1)

wild-type htFdI E15A D7A/D9A/E10A ∆62-64a D7A/D9A/E10A/D28N/E30Q/E32Q I20E Y35E D52R E48A/E49A/D52A E45A/E46A/E48A/E49A/D52A

(3.85 ( 0.38) × 10 (4.17 ( 0.62) × 105 (4.63 ( 0.93) × 105 (1.88 ( 0.19) × 105 (1.19 ( 0.18) × 105 (1.98 ( 0.20) × 105 (1.57 ( 0.31) × 105 (1.04 ( 0.16) × 105 (1.19 ( 0.18) × 105 (1.34 ( 0.13) × 105 (1.04 ( 0.21) × 105 5

∆G° (kJ/mol)

∆H° (kJ/mol)

∆S° (kJ mol-1 K-1)

-31.8 ( 0.3 -32.0 ( 0.4 -32.3 ( 0.5 -30.0 ( 0.3 -28.9 ( 0.4 -30.2 ( 0.2 -29.6 ( 0.5 -28.6 ( 0.4 -28.9 ( 0.4 -29.2 ( 0.3 -28.6 ( 0.5

21.3 ( 1.1 20.4 ( 1.4 13.3 ( 0.9 14.4 ( 1.2 40.1 ( 2.0 7.3 ( 0.7 29.2 ( 1.8 45.4 ( 1.8 32.7 ( 2.0 21.6 ( 1.5 31.2 ( 2.2

0.178 ( 0.004 0.176 ( 0.005 0.153 ( 0.003 0.149 ( 0.004 0.231 ( 0.007 0.126 ( 0.002 0.197 ( 0.006 0.248 ( 0.006 0.207 ( 0.006 0.170 ( 0.006 0.200 ( 0.008

∆62-64 is the form in which the EDE tripeptide of the C-terminal extremity has been deleted.

FIGURE 3: 1H-15N HSQC spectrum of Da htFdI at 298 K. The FdI concentration was 60 µM, in 20 mM phosphate buffer (pH 6.1). Correlation peaks of amide protons are denoted according to the sequence numbering. The magnified peaks represent cross-peaks for I20, G41, and Q44: black for free FdI, blue for FdI in the presence of PFOR (1:0.1), and red for FdI in the presence of PFOR (1:0.2).

We used the eight residues found to be affected by the interaction in the NMR spectrum to define distance constraints to filter the docking results (Figure 5B). Some docked solutions correspond to the chemical shift variations of the eight residues, but the distances between all the PFOR [4Fe4S] centers and the FdI [4Fe-4S] center are larger than 20 Å and, thus, correspond to nonphysiological systems. If a distance filter (FeS-FeS distance of